PAPELES DE TRABAJO EDITADOS POR EL

The evaluation of intestinal microbial diversity by either cultivable or genomic methodologies has been mostly carried out with faecal samples. These samples are considered as representative of the distal large intestine; however, they do not provide knowledge about dynamic microbial process at the site of fermentation in the intestine. All the in vitro fermentative models have the common purpose of cultivate a complex intestinal microbiota under controlled environmental conditions for carrying out microbial modulation and metabolism studies. Thus, they are considered excellent tools to study the human gut microbiota, given the difficulty to access the main niches of colonization in the gastrointestinal tract. Moreover, they allow the screening of a large variety of experimental conditions such as dietary ingredients without ethical constraints.

The simplest in vitro models to study the intestinal microbiota are static batch fermentations. These models are generally performed in small sealed bottles or reactors with either a range of pure or mixed cultures of strains or suspensions of faecal material containing microbial communities from human origin. They are used to describe microbial growth and their ability to metabolize different substrates. These systems have the inconvenient that changes in pH and the accumulation of microbial metabolites lead to the inhibition of microbial activity, so they could not represent the facts occurring in vivo. In spite of these methodological limitations, faecal batch incubations have demonstrated to be a useful tool to investigate metabolic profiles of SCFAs produced by intestinal microbiota from the fermentation of a large variety of dietary components such as inulin-type fructans and resistant starch complex carbohydrates (Pompei et al., 2008; Lesmes et al., 2008). They have

been valuable for a first assessment of intestinal polyphenols metabolism (Gross et al., 2010) or for comparison of different sources or doses of polyphenols (Possemiers et al., 2007; Bolca et al., 2009; De Weirdt et al., 2010). The inoculation density, together with the substrate depletion rate, also defines microbial growth in these systems. Low cell densities allow typical S-shaped growth curve due to the presence of an initial abundant nutrient and the consumption of the substrates at the end of the incubation. Conversely, systems with high cell densities similar to those found in the colon result in limited growth (Payne et al., 2012).

In contrast to short-duration experiments with batch models, long– term experiments with dynamic, multi-compartment gastrointestinal simulators are used when gut microbial dynamics and activities need to be assessed. The clear advantage of these models is to allow real time measurement of the effects of foods or other chemical compounds in the gastrointestinal ecosystem (Kong and Singh, 2010; Wickham and Faulks, 2012). Most of these dynamic models are based on the Reading model firstly described by Gibson and colleagues (1988), which involves a three- stage continuous culture system simulating the ascending, transverse and distal colon. This system allowed setting of pH for the three vessels at 5.5, 6.2 and 6.8 respectively, which is a critical parameter for this microbial ecosystem. The nutritious medium commonly used consists of protein substrates (casein and peptone), complex carbohydrates (pectin, xylan, arabinogalactan and resistant starches) that are not digested by gastrointestinal enzymes, and a mixture of salts and vitamins (Gibson et al., 1988). The continuous replenishment of nutrients and the control of physiological temperature and anaerobic conditions are crucial in the adaptation and survival of the in vitro gut microbiota. This allows the establishment of steady-state conditions in terms of microbial composition and metabolic activity. Furthermore, the control of defined pH values, downstream nutrient limitations and retention times in the different in vitro compartments allows a region-specific differentiation of the microbial

communities and their activity (Van den Abbeele et al., 2010). Anaerobic conditions are usually reached through continuous flushing of CO2 or N2.

Some examples of these dynamic fermentation models are the SHIME (Simulator of the Human Intestinal Microbial Ecosystem), consisting in three glass reactors that represent the ascending, transverse and descending colon and that possess two previous glass vessels representing the gastric and duodenal stages (Molly et al., 1993), and the TIM-2 (Minekus et al., 1999) composed by a continuous single-stage fermenter to simulate the proximal colon conditions. The TIM-2 is able to reproduce the peristaltic mixing of the luminal content and the absorption of water and fermentation products. The previously designed TIM-1 (Minekus et al. 1995) can simulate the process occurring in the upper gastrointestinal tract (stomach and small intestine) and the effluent of the TIM-1 can be manually introduced in the TIM-2 in order to mimic the full transit through the gastrointestinal system (Hatanaka et al., 2012). Comparisons of results obtained from the TIM-2 and SHIME models have demonstrated the suitability of both models when studying the effects of diet on SCFAs production and its influence on specific bacteria (Van den Abbeele et al., 2013). Another example of fermentation model is the twin- vessel single-stage chemostat model recently described by McDonald and colleagues (2014), which reproduces the human distal gut environment by maintaining neutral pH conditions and a constant culture volume. Moreover, recent developments of a single-stage model simulating the ileum microbiota have been reported (Venema and Van den Abbeele, 2013).

The colonization, reproducibility and functional stability of human gut microbiota inside the models depend in part of the inoculum. They can be used diverse alternatives depending on the purpose of the study. Inoculation with liquid faecal suspension from one individual can be useful for biological repetition of a study in order to observe individual variability or stability (Van den Abbeele et al., 2010). Pooling stools from several

individuals can be used to inoculate a batch fermenter and produce a standardized inoculum to be frozen and reused in order to obtain sets of equivalent inocula (Rajilić‐Stojanović et al., 2010; Martinez et al., 2013). Cinquin and colleagues (2004) have developed an immobilization process for the entrapment of faecal microbiota in mixed xanthan-gellan gum gel beads with the purpose of establish biofilm-associated states of microbial populations in conjunction with a continuous wash-out of less competitive bacteria. This last system has recently been updated into the model PolyFermS, allowing maintenance of the microbial diversity over long time and performance of parallel experiments with exactly the same microbiota (Berner et al., 2013). Recent developments focusing in reproducibility of the experiments face the inoculation of the in vitro models with defined populations of pure strains from the human gut microbiota (Newton et al., 2013).

Since most of the models have been focused on simulating either the upper gastric-small intestine digestion or the colonic fermentation process, our research group has pursued the development of a new system which combines the gastrointestinal process from the stomach until the final part of the colon under a unique computer controller. This new system, named SIMGI, is composed by a gastric compartment that simulates peristaltic mixing movements, a reactor simulating the small intestine and three stage continuous fermenters that reproduce the colon region-specific microbiota and its metabolism. It possesses the particularity to allow joint or separated simulation of the gastric and colonic fermentative processes (Barroso et al., 2015).

The ability of the in vitro gut models described above to simulate the in vivo conditions is limited by the lack of the intestinal epithelium and mucus. The evaluation of the host’s response is normally conducted by means of cell culture experiments. They are easy and quick ways to study cellular behaviour, cell signalling pathways and cell interactions. The most

frequent cell lines currently used are Caco-2 and HT29-MTX that are able to form a polarized monolayer of differentiated intestinal epithelial cells (Lesuffleur et al., 1991; Louvard et al., 1992; Sambuy et al., 2005). Furthermore, co-cultures of cell lines in combination with immune cells have been characterized in an attempt to mimic the human intestinal mucosal environment (Christoffersen et al., 2012).

Additional tools for modelling the physiological colonic conditions are the incorporation of mucosal environment devices inside the fermentation vessels in order to differentiate between microbial biofilms adhering to the devices and luminal microbiota (Macfarlane et al., 2005; Van den Abbeele et al., 2012). More recently, the Host-Microbiota Interaction (HMI) module has been developed (Marzorati et al., 2014), consisting of two compartment devices separated by a functional double- layer composed of an upper mucus layer and a lower semi-permeable membrane allowing molecules and oxygen transport. In the upper part, a complex microbial community colonises the mucus, simulating the luminal side of the colon while the lower part contains enterocyte human cells (Marzorati et al., 2014). This combination of in vitro models closely represents the ideal experimental model described by Fritz and colleages (2013) for studying host-microbiota interactions. They stated that physiological pH, fluid retention times and dissolved oxygen concentrations have to be maintained and that the model should include the mucus layer, epithelial cells and complex gut microbiota in anaerobic/microaerophilic conditions.

Overall, to get the complete picture, combining both in vitro and in

vivo studies could represent an appropriate approach to understand how

microbiota interacts with the host. The use of animal models is an excellent tool for this purpose. For example, diet-induced alterations in microbiota composition of zebrafish have shown to influence fat absorption (Semova et al., 2012). Moreover, development of germ-free

mice (Reyniers et al., 1959) has helped to understand the microbiota-host interactions and have allowed highly controlled working conditions. Transplantation of human faecal microbiota into germ-free mice can be viewed as capturing an individual’s microbial community at a fixed moment in time (Goodman et al., 2011; Kau et al., 2011; Ridaura et al., 2013).

Nevertheless, clinical trials are required to confirm results obtained through all of these study models. As it has been exemplified above, until date, several human trials have been performed in order to test the microbiota modulatory effects of probiotics (Savard et al., 2011; Johnston et al., 2011), prebiotics (Roberfroid et al., 2010) and other components of the diet such as polyphenols (Etxeberria et al., 2013).

The growing evidence about the important role of the human microbiota in health leads to the development of new dietary ingredients and approaches for modulating the microbiota. Thus, it is necessary to perform both in vitro and in vivo experiments to deeper understand the dynamic interactions inside the human microbial community and with the host. Further, it remains essential to conduct more studies combining biochemical, microbiological and clinical parameters in order to know the effectiveness of the microbial modulatory strategies and their impact on health.